Crack Detection using Edge Detection and Transfer Learning Models

Chopra, Pooja (2022) Crack Detection using Edge Detection and Transfer Learning Models. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration manual) Download (1MB) | Preview

Abstract

Cracks in bridges can cause severe loss of life, money and property. Early detection and continuous monitoring can avoid the collapsing of bridges. The manual inspection of cracks requires specialist experience, and it is a tedious task that requires plenty of time. This paper gives an alternative approach to monitoring cracks in the bridge. Images collected from the drone are used in this study for training the network for detecting cracks. The cracks in the bridge’s deck, walls and pavements are detected individually and together using transfer learning algorithms such as xception, resnet50 and VGG19. These three models were compared for this task, and it was found that xception model outperformed these three models. 2 extra ReLu layers were added to xception model to increase its efficiency of the model. A few transformations were carried out on images, such as image segmentation, canny transformation and changing the image’s contrast. These transformations increased the accuracy of models on deck and pavement.

Item Type:	Thesis (Masters)
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science T Technology > TG Bridge engineering T Technology > TR Photography
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Tamara Malone
Date Deposited:	19 Jan 2023 16:58
Last Modified:	19 Jan 2023 16:58
URI:	https://norma.ncirl.ie/id/eprint/6103

Actions (login required)

View Item